Add qsat to restart files to enable warm starts with deposition#304
Add qsat to restart files to enable warm starts with deposition#304LeonGeers wants to merge 1291 commits into
Conversation
Not used anymore
* modforces: account for gravity on precipitation with bulkmicro3 * modforces: change OpenACC kernels to parallel loop --------- Co-authored-by: Caspar Jungbacker <caspar.jungbacker@outlook.com>
* modsurface: change lmostlocal to .true. by default reason: with large domains with some heterogeneity, e.g. with open boundaries or cold pools, there may be large deviations in the local wind speed from the domain average. Using the local wind by default is less surprising than using the domain average. * modsurfdata: add space :-) --------- Co-authored-by: Caspar Jungbacker <74240267+CasparJungbacker@users.noreply.github.com>
modchem: avoid out-of-bounds read when determining lastchem, fix typos
* Nudging: add support for nudging scalars * Tracers: add switch for enabling nudging * Nudging: small fixes for nudging scalars
Aerosols Infrastructure for M7 aerosol scheme. Based on an earlier implementation by Marco de Bruine. This version is very bare bones, since it only features activation. More features to come...
documentation updates
* Poisson: move reading of solver namelist * Poisson: move data to `modpois_data.f90` * Poisson: re-add default solver selection * Poisson: actually track modpois_data.f90
modchecksim: fix stars in sim_sec/s (add min/hour option)
* modnudge: move printing of nudging profile to the file used_profiles.txt * update copyright years, remove old changelog * program.f90: remove duplicate call to datetime
baseprof: re-add user-specified baseprof option
Refactor and simplify timers
Fix fortran-support compilation with `-fdefault-real-8`
|
Good find, let's fix this. However, this pull request is massive because it tries to merge into the main branch. Probably it should target dev, with just the one cherry-picked commit. Second, qsat can be calculated from the prognostic variables, so it might be possible to not store it in the restart file but make sure it's been calculated in the time step before it's required - that's nice also otherwise because then the qsat value used is the most recent one. Not sure if this is easy or possible in this case. |
|
Oh, crap. I was under the impression that, since it is a cherry-pick to a sidebranch of dev, it should generate a merge request in dev. F'd it up again, apparently. I'll have a look at it tomorrow |
|
qsat is calculated in every time step, a fix that was made some time ago, which renders this merge request redundant. |
8 participants
Ruud found a problem running warm starts with DALES, specifically when deposition was activated. It turned out that the compensation point concentration was NaN, caused by the fact that qsat was not defined when restarting (it was not included in the warm start initializer files). He's corrected that for DALES v4.4 (Ruisdael branch) and I cherry-picked it for version 5.